STATUS REPORT 19 th North America/Europe Data Exchange Meeting NOAA Silver Spring, MD May 3-5, 2006 Paul Poli (CNRM/GMAP) replacing Bruno Lacroix (DPrévi/COMPAS) With contributions from CNRM/GMAP and
Jan 19, 2016
STATUS REPORT
19th North America/Europe Data Exchange Meeting NOAA Silver Spring, MD May 3-5, 2006
Paul Poli (CNRM/GMAP)
replacing Bruno Lacroix (DPrévi/COMPAS)
With contributions from CNRM/GMAP and CNRM/COMPAS
2Outline
• Introduction
• Operational suite(s)– Update on current configurations
– New: 3DVAR assimilation in LAM, chemical transport model
• Test suite– NOAA-18 AMSU-A+MHS
– Aqua/Terra MODIS winds
– GOES/MTSAT cloud-track winds BUFR
• Issues under development– SSM/I
– GPS ZTD
• Future Plans
3Introduction : Ballpark Figures
• Total number of observation counts assimilated every 6 hours in Meteo France global model: 260,000~325,000
(Out of total rcvd)– SYNOP&BUOY (95%)– TEMP (82%)– AIREP (55%)– Wind Profilers (21%)– Geostat. Winds (7%)– ATOVS (4%)– Quikscat (28%)
“Conventional”
Satellite
40% to 60%of total obs used
00,12 UTC 06,18 UTC
4Assimilation/Forecast Suites
• Computing platform– Since mid-2003: Fujitsu VPP 5000, 124 PE
– 2 machines: operations (60 PEs) and research/backup (64 PEs)
– Replacement expected 2006 Q4
• Operational suites:– Global stretched
– Global stretched, very short cutoff (00 UTC only)
– Global uniform
– Limited-Area
– Ensemble Forecast
• Data monitoringhttp://www.meteo.fr/special/minisites/monitoring/menu.html
User/password available upon request [email protected]
5Each Assimilation/Forecast Suite
• Two cycles running in parallel– ‘Production cycle’:
• Short cutoff
• Long forecasts
– ‘Assimilation cycle’:• Longer cutoff
• Forecast only +6 hours (first-guess for production cycle)
• Keeps the “memory” of the assimilation system
6
Stretched Global Model Horizontal Resolution
23km
133km133km
7
Supercomputer Platform Machine Load due to Operations only
Global stretched model: 4DVAR analysis in 30 minutes, 24H forecast in 20 minutes
8
Modifications in Operational Suites since Last Data Exchange Meeting (April 2005)
• June 2005: Start of a chemical transport suite to issue air quality forecasts
• July 2005: Start of a 3DVAR assimilation for Limited-Area Model (LAM) over Europe
• January 2006: RS bias correction in global model
9
Chemical Transport Model Operational since 27/06/2005
• 3 domains:– Global/Europe/France
– Horz. resolutions: 4°x4°, 0.5°x0.5°, 0.1°x0.1°
• Observations– Observations currently only used for validation
– Plans to move to assimilation
• Forecasts of air quality up to 96H
10
3DVAR Limited Area Model Operational since 25/07/2005
• Assimilates same observations as operational global model– With the addition of
• Meteosat-8 clear-sky IR radiances from instrument SEVIRI
• T2m and q2m
– Without:• Quikscat
– Only over Europe
– Using 3DVAR
T. Montmerle
11
72H Forecast RMS Error of Global Models over Europe as Compared to RS
Stretched
Uniform
NOAA-16 AMSU-B, Quikscat, EARS
NOAA-17 NCEP SST, sea-ice mask from SSM/I
HIRS, high frequency SATOB
Aqua AMSU-A
12Ocean Wave Models
• 4 models:– Global stretched & uniform, Europe, France
– Horz. resolutions: 1°, 1°, 0.25°, 0.1°
– Global stretched & uniform models assimilate Jason-1 and Envisat altimeter wave height data
– Forecasts up to 102H, 72H, 54H, 54H
J.M. Lefevre
13Observations Usage Summary
Datatype Contact Operations Test Suite
ATOVS E. Gerard
Geostat. winds C. Payan
MODIS winds P. Moll
Quikscat winds C. Payan
Conv. (RS, AIREP, TEMP) P. Moll
GPS ZTD P. Poli (monitoring)
AIRS F. Rabier (monitoring)
N15,16,17,Aqua
GOES,MeteoSat, MTSAT
N18
Aqua,Terra
14Passive Sounders Usage Summary
Satellite/Instrument
NOAA-15
NOAA-16
NOAA-17
NOAA-18
Aqua
AMSU-A Oper Oper Test Oper
AMSU-B or MHS Oper Oper Test
HIRS Monitoring Oper
AIRS Monitoring
15Current Test Suite
Changes as compared to operational suite:
• 46 model levels instead of 41
• Model top 0.1hPa instead of 1hPa
• New radiation scheme
• New physics scheme; now includes following prognostic variables– Cloud liquid water: in suspension, and falling
– Cloud ice water: in suspension, and falling
• Addition of NOAA-18 AMSU-A and MHS
• Cloud-track winds from GOES and MTSAT now in BUFR
• Addition of GPS ZTD monitoring over Europe
16
Current Coverage in Operations of NOAA/NASA AMSU-A Production cycle
17
Current Coverage in Operations of NOAA/NASA AMSU-A Assimilation cycle
18
Test suite Coverage for NOAA/NASA AMSU-A Assimilation cycleAddition of NOAA-18
19
Current Coverage in Operational suite of Cloud-track Winds
20
Test Suite Coverage for Cloud-track Winds
Addition of Aqua/Terra MODIS winds GOES&MTSAT-1R winds now BUFR
21SSM/I Tests with only F13, F15
E. Gérard
• Over oceans only
• 250 km thinning
• SSM/I-specific quality control:
–Cloudy areas (CLWP > 0.1 kg.m-2)
–Rainy areas (BT37V - BT37H < 40 K)
22
SSM/I Tests: Impact on AnalysisTotal Column Water Vapor
First guessAnalysis
SSM/I assimilation adds water in the tropics – but later
increments stable
With SSM/I
Without SSM/I
E. Gérard
23
SSM/I Tests: Impact on AnalysisTotal Column Water Vapor
Analysis with SSM/I
Increments in SSM/I
experiment
Increments in control
experiment
Difference (analysis with SSM/I) minus
(control analysis)
E. Gérard
24
SSM/I Tests: Impact on Analysis Low Cloud Cover Difference
More clouds with SSM/I
Less clouds with SSM/I
E. Gérard
25
SSM/I Tests: Impact on Forecast Geopotential Scores w.r.t. RS
RMS Std. Dev. Bias
Blue = Improvement Red = DegradationE. Gérard
26GPS ZTD Tests
1 mm ZTD approx. 0.5 hPa surface pressure
1 mm ZTD approx. 0.15 mm integrated water content
er transmittGPS
receiver GPS
1ZTD dssn
GPSTransmitter
ground-based GPS receiver
rad
io link
More than 500 stations over Europe, 11 processing centers more than 900 distinct data sources
Atmospheric refractive index
27GPS ZTD Tests
Station selection process
28
GPS ZTD Tests: Fit of ZTD Data by Processing Center w.r.t. Global Model
Processing center with stations located
in mountaineous area
29GPS ZTD Tests: 4DVAR Assimilation
time
1 2 3 4 5 6 7
Predicted model trajectory (forecast)
Analysis
Observations
Timeslots
-2H30-3H00
-1H30-0H30
+0H30+1H30
+2H30+3H00
30GPS ZTD Tests
• Fixed station selection• Fixed bias correction
• Automatic station selection• Sliding average bias correction
Blue = Improvement Red = Degradation
Blue = Improvement Red = Degradation
31
GPS ZTD Tests: Quantitative Precipitation Forecasts over France (36H-
12H)Frequency Bias Index
Equitable Threat Score
Probability of Detection
False Alarm Rate
WITH GPS
WITH GPS
WITHOUT GPS
WITHOUT GPS
32
Radiosonde Minus Analysis Temperature Bias [contour = 1K]
~1K at 150hPa
Switch to raw radiances
33Plans for the Coming Year
• 4DVAR assimilation in chemical transport model in test suite
• GPS ZTD assimilation in operations (global model)
• SSM/I assimilation in operations• AIRS assimilation in operations• GPS RO in test suite• Limited-area model over La Réunion
• Transition of all suites to new platform
! THANK YOU FOR YOUR ATTENTION !
19th North America/Europe Data Exchange Meeting NOAA Silver Spring, MD May 3-5, 2006